Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

Quantum Mechanics 2
Dr Juan Rojo
VU Amsterdam and Nikhef Theory Group
http://www.juanrojo.com/ , j.rojo@vu.nl
Current version: February 23, 2021

3 Chapter 5: Identical Particles in Quantum Mechanics

Learning Goals

• To formulate and solve the Schroedinger equation in the case of two-particle systems.

• To determine the implications that in quantum mechanics we deal with systems of genuinely
indistinguishable particles.

• To formulate the Pauli exclusion principle and to derive its consequences for the structure
of multi-electron atoms.

• To explain fundamental properties about the general structure of solids by means of basic
quantum-mechanical properties.

• To formulate the concept of band structure in solids and use it to classify materials in terms
of their electrical conductivity properties.

In this section of the lecture notes we present the main concepts discussed in Chapter 5 (“Identical Par-
ticles”) of the course textbook. The goal of these lecture notes is to provide a self-consistent study resource
for the students, which is then complemented by the live lectures (and their recordings), the tutorial sessions,
as well as their own study of the textbook. The relevant textbook sections are indicated below, material
from other sections not listed there will not be required for the examination.

Textbook sections
• 5.1: Two-Particle Systems.

• 5.2: Atoms.

• 5.3: Solids.

Page 55 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

Up to now, in your study of quantum mechanics you have only considered one-particle systems (including
the hydrogen atom, which can be expressed essentially as a one-particle problem). As you know from classical
mechanics, in general the description of many-body systems ramps up quickly in difficulty: the two-body
problem in classical mechanics can be solved exactly but already the three-body problem does not admit a
closed solution. However, at the classical level, systems of non-interacting particles are trivial, in that
their solution is the combination of independent solutions of the one-particle problem.
However, the situation is completely di↵erent in the case of quantum mechanics, and the state vectors
for systems composed by more than one particle will in general be di↵erent from a trivial combination of
one-particle states. The reason for this marked di↵erence is that in quantum mechanics we often deal with
systems of indistinguishable particles: an electron is identical to any other electron anywhere in the
Universe.
This appears at first sight to be a rather minor point, but careful consideration of the implications that
dealing with identical particles has in quantum mechanics has deep consequences, from explaining what
makes a solid well, “solid”, to the allowed options in which spins can be combined in many-particle states
(which has impact in for instance in quantum technologies). And these implications are already there even
for non-interacting particles: as we will see we can obtain many interesting results even neglecting the
interactions between the particles in our system.

3.1 Two-particle systems

As mentioned above, up to now we have considered quantum systems composed by a single particle. For
example, we have solved the Schroedinger equation in three dimensions for the hydrogen atom and found
the wave functions of the electronic orbitals nlm (r, ✓, ). Note that although in this case the system was
composed by a proton and an electron, the proton did not play any role other than generating the electric
potential thus was not part of the wave function of the system (technically, it did not play any dynamical
role in the system).
We will consider now a quantum state composed by two dynamical particles. Its wave function will
now depend on the position of the two particles, (r1 , r2 , t) and its time evolution will be determined by
the solution of the time-dependent Schroedinger equation
✓ ◆
@ (r1 , r2 , t) ~2 2 ~2 2
i~ = Ĥ (r1 , r2 , t) = r r + V (r1 , r2 , t) (r1 , r2 , t) , (3.1)
@t 2m1 1 2m2 2

where in general the two particles will have di↵erent masses m1 and m2 , and the potential energy now
depends on the position vectors of the two particles as well as on the time, V (r1 , r2 , t). The generalised
statistical interpretation tells us that the square of the wave function, | (r1 , r2 , t)|2 , will be normalised to
one when integrating over all possible values of r1 and r2 .
Here we will consider time-independent potentials, such that the time evolution of the eigenfunctions of
the Hamiltonian is given as usual by
✓ ◆
iEt
(r1 , r2 , t) = (r1 , r2 ) exp , (3.2)
~

where the spatial component of the wave function satisfies the time-independent Schroedinger equation
✓ ◆
~2 2 ~2 2
r r + V (r1 , r2 ) (r1 , r2 ) = E (r1 , r2 ) , (3.3)
2m1 1 2m2 2

Page 56 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

in terms of the kinetic terms for the two particles and of the potential energy. Solving a two-body Schroedinger
equation is in general much more difficult that its one-body counterparts, but in some cases we actually can
reduce the solution of Eq. (3.3) to the solutions of separate one-body problems.
The simplest scenario for Eq. (3.3) is that of non-interacting particles, where each particle moves
in some external potential but they do not interact among them. Note that here the meaning of “non-
interacting” is not that we have two free particles, but that the two particles in the system interact only
with an external potential but not among them. In this case, the potential energy simplifies to

V (r1 , r2 ) = V1 (r1 ) + V2 (r2 ) , (3.4)

and it is easy to see that Eq. (3.3) admits a solution where the total wave function factorises into two
one-particle wave functions
(r1 , r2 ) = a (r1 ) b (r2 ) , (3.5)

where a and b indicate two di↵erent one-particle wavefunctions, each of which satisfies their a separate
one-particle Schroedinger equation:
✓ ◆
~2 2
r + V1 (r1 ) a (r1 ) = Ea a (r1 ) , (3.6)
2m1 1
✓ ◆
~2 2
r + V2 (r2 ) b (r2 ) = Eb b (r2 ) ,
2m2 2

with the total energy of the system being given by E = Ea + Eb . Note that the situation where we have
two free particles is a particular case of this, and the equations that we need to solve are:

~2 2
r a (r1 ) = E1 a (r1 ) , (3.7)
2m1 1
~2 2
r b (r2 ) = E2 b (r2 ) .
2m2 2

Note, however. that if we know the solutions of the one-particle equations, Eq. (3.6), then the case of two
particles interacting in some external potential falls on the same complexity class as that of a system com-
posed by two free particles, and it is not more difficult to solve.

Entanglement in two-particle systems

For non-interacting particles, the two-body wave function factorises into two one-body wave functions
of the form
(r1 , r2 ) = a (r1 ) b (r2 ) ,

which are eigenvalues of the total Hamiltonian Ĥ with energy E = Ea + Eb . However, this does not
imply that all valid quantum states of this two particle system can be described by such factorisable
state vectors. For example, if c and d are two other one-particle solutions, surely I can have
r r
1 2
(r1 , r2 ) = a (r1 ) b (r2 ) + c (r1 ) d (r2 ) (3.8)
3 3

which cannot be expressed as a product of one-particle wavefunctions. These states are

called entangled states and are of huge importance in quantum mechanics.

Page 57 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

Another situation that admits a relatively compact analytic treatment is that of central potentials, such
as the Coulomb potential that we discussed in the case of the hydrogen atom. In this case, the interaction
between the two particles depends only on their relative distance but not on their relative orientation

V (r1 , r2 ) = V (|r1 r2 |) . (3.9)

Then the two-body Schroedinger equation

✓ ◆
~2 2 ~2 2
r r + V (|r1 r2 |) (r1 , r2 ) = E (r1 , r2 ) , (3.10)
2m1 1 2m2 2

can be factorised into two separate one-body equations, one for the motion around the center of mass
of the system and the other for the free motion of the center of mass.
Thus we have found that there are two situations where solving a two-particle Schroedinger equation is
relatively feasible:

• When the two particles do not interact with each other, but only with some external potential.

• When the two particles interact with each other via a central potential, but then do not interact
with any external potential.

In both cases, we can reduce the two-body system to two one-particle problems that can be solved separately.
However, in general the two particles will experience both interactions with some external force and mutual
interactions between them. An important example in this context is the helium atom, a system composed
by an atomic nucleus with two protons Z = 2 and then two electrons orbiting around it. The potential
experienced by the electrons receives contributions both from the electrical attraction with the positively
charged nucleus and from the negative repulsion between the two electrons:
✓ ◆
1 2e2 2e2 e2
V (r1 , r2 ) = + . (3.11)
4⇡✏0 |r1 | |r2 |r1 r2 |

The Schroedinger equation for the Helium atom cannot be solved analytically in closed form. We will need
to introduce some approximations such as perturbation theory and the variational principle, which we will
introduce later in the course.
Now, as we discuss next, even for the cases where we can solve exactly the Schroedinger equation for the
two body problem (say for non-interacting particles), there remain some important issues to be addressed
in the case that the two particles that compose the system are indistinguishable. To illustrate this point,
when we wrote the factorised solution for the total wave function of a system composed by two non-interacting
particles we had
(r1 , r2 ) = a (r1 ) b (r2 ) , (3.12)

we implicitely assumed that we could distinguish the two particles, since we assume that the first particle
occupies the one-particle state a and the second b . But this is not possible if I can tell apart the two
particles! We will see next how to quantum mechanics solves this conundrum.

Indistinguishable particles: fermions and bosons Let us assume that we have now two noninteracting
particles, the first one occupying the one-particle state a (r) and the second one in another one-particle state
10
b (r). As demonstrated above, the total wave function of the two-particle system will be the product of
10 While here I work with position-representation wavefunctions, the concept is fully general and can also be formulated in

terms of the Dirac notation.

Page 58 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

the two one-particle states:

(r1 , r2 ) = a (r1 ) b (r2 ) . (3.13)

Now, by writing the total wave function in this manner I am assuming that somehow we can tell apart
the “first particle” from “the second particle” (that is, that we can stick labels on them). But in quantum
mechanics particles are indistinguishable, not because of some technical limitation of our measurement
apparatus, but because every electron is identical to every other electron, and every hydrogen atom
is identical (and will always be) identical to any other hydrogen atom. In other words, in quantum mechanics
it is not possible to say “this electron” or “that electron”, but only “a electron”.
To take into account that quantum mechanics deals with particles that are indistinguishable, we need
to modify the two-body wavefunction for identical particles in a way that it does not implicitely assume
that we can tell particles apart. This can de done in two ways. First, we can write

+ (r1 , r2 ) = A[ a (r1 ) b (r2 ) + b (r1 ) a (r2 )] . (3.14)

where the only thing that we assume is that there is “a particle” in orbital a and another of the same
particles in orbital b , and where A is an overall normalisation constant. Another possible option that
achieves the same goal is
(r1 , r2 ) = A [ a (r1 ) b (r2 ) b (r1 ) a (r2 )] . (3.15)

with the minus sign between the two terms.11

Actually, given a specific type of particle, only one of these two relations will be valid. If we assume
normalised one-particle states, then we can have only one of the following two situations:

• We denote as bosons as those elementary particles for which their two-body wave function is

1
p [ a (r1 ) b (r2 ) + b (r1 ) a (r2 )] .
2

Bosons have the defining property that their total wave function is symmetric under the exchange
of the two particles: + (r1 , r2 ) = + (r2 , r1 ). All particles with integer spin are bosons (for example
the photon, with s = 1 or the Higgs boson, with s = 0).

• We denote as fermions those elementary particles for which their two-body wave function is

1
p [ a (r1 ) b (r2 ) + b (r1 ) a (r2 )] .
2

Fermions have the defining property that their total wave function is antisymmetric under the ex-
change of the two particles : (r1 , r2 ) = (r2 , r1 ). All particles with half-integer spin are
fermions (for example the electron and the proton, both with s = 1/2).

Here we take this property as an axiom of quantum mechanics. The formal connection between the spin
of a particle and the symmetry properties of the total wave function under the exchange of two particles can
be demonstrated in quantum field theory and is known as the spin-statistics theorem. Here the situation
is similar as for the Heisenberg uncertainty principle, which despite its name was not a fundamental principle
but rather a consequence of the more fundamental axioms of quantum mechanics.
11 Actually,there are cases where one can have also a complex phase between the two terms, leading to exotic quasi-particles
such as anyons, but we will not discuss this case further in this course.

Page 59 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

The Pauli exclusion principle

A direct consequence of the of the spin-statistics axiom of quantum mechanics is that two
fermions cannot occupy the same quantum state. To see this, consider two non-interacting
fermions that occupy the same one-particle quantum state a . The combined two particle wave
function would then be:
1
(r1 , r2 ) = p [ a (r1 ) a (r2 ) a (r1 ) a (r2 )] = 0. (3.16)
2

hence only when the two one-particle states are di↵erent from each other the two-particle wave
function of a fermion system will be non-zero. This central result of quantum theory is known as the
Pauli exclusion principle and as we will see later in this chapter has crucial implications for the
structure of matter.

The infinite square well with two particles. In order to illustrate how di↵erences between fermions
and bosons can have very significative consequences for the properties of quantum systems, let us consider
the following system. Assume that we have two noninteracting particles with the same mass confined
in the infinite square well potential in one dimension. For simplicity, let’s work in units where the width of
the well is L = 1 and that the particle mass is m = 1. The potential energy for this system is then

V (x) = 0 0 < x < 1,

V (x) = 1 x  0,x 1.

The one-particle states and energies of this potential are are quantised in terms of a single principle quantum
number n and are given by

p n2 ⇡ 2 ~2
n (x) = 2 sin(n⇡x) , En = , n = 1, 2, 3, . . . . (3.17)
2
Let us consider three di↵erent scenarios for this system, and as we will see whether or not particles are
distinguishable or indistinguishable, and in the latter case whether they are fermions or bosons, has important
consequences for the resulting energy spectrum of this system.

• The two particles are distinguishable, that is, they are not identical.
If we call n1 the energy level occupied by the first particle and n2 that by the second particle, the total
wave function is just the regular product of one-particle wave functions

(n21 + n22 )⇡ 2 ~2
n1 n2 (x) = n1 (x1 ) n2 (x2 ) , En1 n2 = En1 + En2 = ⌘ (n21 + n22 )K . (3.18)
2
For example, the ground state is 11 (x) with energy E11 = 2K and the first excited state is double-
degenerate, with two possible two-particle states having associated the same total energy

21 (x1 , x2 ) = 2 sin(2⇡x1 ) sin(⇡x2 ) , E21 = 5K ,

12 (x1 , x2 ) = 2 sin(⇡x1 ) sin(2⇡x2 ) , E12 = 5K . (3.19)

By varying n1 and n2 , we can then construct all the possible two-particle wave functions allowed for
this system.

Page 60 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

Note that each two-particle state is uniquely specified by the choice of the two one-particle quantum
numbers n1 and n2 . In this respect, the two-particle states are trivial combinations of one-particle
states, much in the same way as what would happen in classical physics for non-interacting particles.

• The two particles are indistinguishable, that is, they are identical, and specifically they are bosons.12
In this case, the ground state will be unchanged, since clearly the two-particle wave function

11 (x1 , x2 ) = 2 sin(⇡x1 ) sin(⇡x2 ) , (3.20)

is symmetric under the exchange of the two particles:

11 (x2 , x1 ) = 2 sin(⇡x2 ) sin(⇡x1 ) = 11 (x1 , x2 ) , (3.21)

as should be the case for the wave function of a system composed by bosons.
However, you also notice that the (degenerate) first excited states in Eq. (3.19) are not symmetric
under the exchange of the two particles. For example, if I take 21 and I exchange the particles

21 (x2 , x1 ) = 2 sin(2⇡x2 ) sin(⇡x1 ) = 12 (x1 , x2 ) 6= 21 (x1 , x2 ) , (3.22)

and hence 21 (x2 , x1 ) and 12 (x2 , x1 ) constructed as in Eq. (3.19) are not acceptable for a system
composed by two identical bosons.
Instead, we must symmetrise the two-particle wave function corresponding to the first excited state of
the system by using the appropriate prescription for bosons. In this way we find that the wave function
of the first excited state is given by
p
(x1 , x2 ) = 2 [sin(2⇡x1 ) sin(⇡x2 ) + sin(⇡x1 ) sin(2⇡x2 )] , E = 5K , (3.23)

which is obviously invariant under the exchange of the two particles, (x2 , x1 ) = (x1 , x2 ).
This result illustrates a first important consequence of the quantum statistics: the first excited state,
which is degenerate in the case of distinguishable particles, becomes non-degenerate in the case of
indistinguishable bosons. That is, quantum statistics breaks the degeneracy which is present in the
case of distinguishable particles.

• The two particles are indistinguishable, that is, they are identical, and specifically they are fermions.
In this case the two-particle state which corresponding to the ground state of the previous two cases is
not allowed due to the Pauli exclusion principle: two fermions cannot occupy the same one-particle
quantum state. In this case the wave function of the ground state (the state with the lowest energy) is
p
(x1 , x2 ) = 2 [sin(2⇡x1 ) sin(⇡x2 ) sin(⇡x1 ) sin(2⇡x2 )] , E = 5K , (3.24)

which as requested for fermion systems is antisymmetric under the exchange of any two particles,
given that (x1 , x2 ) = (x2 , x1 ).
Here we note another very powerful consequence of quantum statistics: as compared to the case with
distinguishable particles, or to the case with indistinguishable bosons, in the case of fermions the
12 Note that the following discussion requires the two particles to be identical bosons, not only bosons. For example, a system

composed by two distinct spin-1 particles can still be treated as in the previous case, since here we do not deal with genuinely
indistinguishable particles.

Page 61 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

energy of the ground state is increased. In other words, keeping everything the same (mass,
potential etc) the zero-point energy of this quantum system is increases by a factor 2.5 just from
replacing bosons by fermions, which is quite a remarkable result.

Hence, to summarize we have illustrated with this relatively simple example some important consequences
of accounting for the fact that quantum mechanics deals with indistinguishable particles: as compared to
system composed by distinguishable particles, we can

• Change the energy and eigenfunctions spectrum of the system by removing degeneracies.

• Change the energy and eigenfunctions spectrum of the system by increasing the zero-point energy
of the ground state.

In the rest of this chapter, we will study other implications that systems of indistinguishable particles have
in quantum mechanics.

Exchange forces. The previous example demonstrates that the quantum statistics that apply in systems
composed by identical particles can lead to significance consequences in the properties of quantum states.
Let us investigate a bit more what is the underlying reason for this behaviour.
Consider now a one-dimensional system composed by two particles, and consider two one-particle quan-
tum states a (x) and b (x) (orthonormal between them). Following the previous discussion, the two-particle
wave function will be given in the various relevant cases by:

• Distinguishable particles: (x1 , x2 ) = a (x1 ) b (x2 ).

• Indistinguishable particles (bosons): + (x1 , x2 ) = p1

2
( a (x1 ) b (x2 ) + b (x1 ) a (x2 )).

• Indistinguishable particles (fermions): (x1 , x2 ) = p1

2
( a (x1 ) b (x2 ) b (x1 ) a (x2 )).

Let us now evaluate the expectation value of the square of the separation between the two particles:
⌦ ↵ ⌦ ↵ ⌦ ↵
(x1 x2 )2 = x21 + x22 2 hx1 x2 i , (3.25)

in the three cases under consideration, and compare their results. The reason to choose this quantity is
that, as we will demonstrate, quantum statistics leads to an e↵ect on the average distance between the two
particles which is akin to either a repulsive or attractive force, known as exchange forces. However, this
term is a bit of a misnomer since the particles are non-interacting: it will look like there is a force acting
upon them, but it is just a geometrical consequence of how wave functions of identical particles must be
constructed in quantum mechanics.

• Distinguishable particles: in this case the relevant expectation values are

Z Z Z Z
⌦ 2↵ 2 2
⌦ ↵
x1 = dx1 dx2 x1 | (x1 , x2 )| = dx1 x1 | a (x1 )| dx2 | b (x2 )|2 = x2 a , (3.26)

where we have used that the one-particle wave functions are appropriately normalised. The result is
then the expectation value of the position squared in the one-particle state a (x). Using a similar
strategy, we can easily evaluate the other matrix elements to find:
⌦ ↵ ⌦ ↵
x22 = x2 b , hx1 x2 i = hxia hxib , (3.27)

Page 62 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

and hence the sought-for expectation value for the square of the distance between the two particles in
the system is given by
⌦ ↵ ⌦ ↵ ⌦ ↵
(x1 x2 )2 = x2 a + x2 b 2 hxia hxib . (3.28)
⌦ ↵
Note how this result is consistent with the particles being distinguishable: (x1 x2 )2 is determined
by combining matrix elements that only involve one-particle states: the two particles don’t talk to
each other at all ( as should be the case, since these are non-interacting particles).
Let’s compare this result with the other two cases, corresponding now to indistinguishable particles.

• Indistinguishable particles: now the same calculation yields

Z Z Z Z
⌦ 2↵ 2 2 1
x1 = dx1 dx2 x1 | ± (x1 , x2 )| = dx1 dx2 x21 | ( a (x1 ) b (x2 ) ± b (x1 ) a (x2 )) |
2
2
Z Z
1
= dx1 dx2 x21 | a (x1 ) b (x2 )|2 + | b (x1 ) a (x2 )|2
2
!
⇤ ⇤ ⇤ ⇤
± a (x1 ) b (x2 ) b (x1 ) a (x2 ) ± b (x1 ) a (x2 ) a (x1 ) b (x2 )

Z Z Z Z
1 1
= dx1 x21 | a (x1 )|2 dx2 | b (x2 )|2 + dx1 x21 | b (x1 )|2 dx2 | a (x2 )|2 (3.29)
2 2
Z Z
1
± dx1 x21 a⇤ (x1 ) b (x1 ) dx2 b⇤ (x2 ) a (x2 )
2
Z Z
1 1 ⌦ 2↵ ⌦ ↵
± dx1 x21 b⇤ (x1 ) a (x1 ) dx2 a⇤ (x2 ) b (x2 ) = x a + x2 b ,
2 2

where we have used that the one-particle states are normalised and orthogonal among them. Clearly,
⌦ ↵
we will get the same result for x22 , since the two particles are indistinguishable. Now for the third
term in the calculation, it is when things become interesting:
Z Z
hx1 x2 i = dx1 dx2 x1 x2 | ± (x1 , x2 )|2
Z Z
1
= dx1 dx2 x1 x2 | ( a (x1 ) b (x2 ) ± b (x1 ) a (x2 )) |2
2
Z Z
1
= dx1 dx2 x1 x2 | a (x1 ) b (x2 )|2 + | b (x1 ) a (x2 )|2
2
!
⇤ ⇤ ⇤ ⇤
± a (x1 ) b (x2 ) b (x1 ) a (x2 ) ± b (x1 ) a (x2 ) a (x1 ) b (x2 )

which by evaluating each of the individual integrals gives

Z Z
1
hx1 x2 i = dx1 x1 | a (x1 )|2 dx2 x2 | b (x2 )|2
2
Z Z
1 2
+ dx1 x1 | b (x1 )| dx2 x2 | a (x2 )|2 (3.30)
2
Z Z
1
± dx1 x1 a⇤ (x1 ) b (x1 ) dx2 x2 b⇤ (x2 ) a (x2 )
2
Z Z
1
± dx1 x1 b (x1 ) a (x1 ) dx2 x2 a⇤ (x2 ) b (x2 )
⇤
2
1
= (2 hxia hxib ± hxiab hxiba ± hxiba hxiab ) = hxia hxib ± | hxiab |2 ,
2

Page 63 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

in terms of the overlap integral between the two one-particle states defined as
Z
hxiab ⌘ dx x a⇤ (x) b (x) . (3.31)

Note that this overlap integral arises only when we have indistinguishable particles: it it a conse-
quence of the cross-terms that we get when we square the (anti-)symmetric two-particle wave function
± (x1 , x2 ) (so it is a direct consequence of the quantum statistics requirements).

Putting together the three terms, we obtain the counterpart of Eq. (3.28) for indistinguishable particles:
⌦ ↵ ⌦ ↵ ⌦ ↵
(x1 x2 ) 2 = x2 a + x2 b 2 hxia hxib ⌥ 2| hxiab |2 , (3.32)

which is di↵erent from the case of the system composed by two distinguishable particles, and further-
more the result varies if we consider a system of bosons and a system of fermions.

Combining these results together, we therefore obtain that the distinguishability or not of the particles
that compose a two-particle quantum system has a direct physical consequence in the expectation value of
the square of their average spatial distance:
D E D E
2 2
( x) = ( x) ⌥ | hxiab |2 , (3.33)
± d

which is di↵erent for bosons than for fermions:

• For identical bosons, on average the two particles will tend to be closer together as compared to the
case of two distinguishable particles.

• For identical fermions, on average the two particles will tend to be farther apart as compared to the
case of two distinguishable particles.

The exchange force for identical particles

The result of Eq. (3.33) indicates that a system of two identical bosons behaves as if there was an
attractive force between then, while a system of two identical fermions behaves on the contrary as
if there was a repulsive force between then. This putative force is often called exchange force,
which is not really accurate since there is no physical interactions between the two particles (the
system is non-interacting!), it is a consequence of the symmetry requirements on the two-particle
state vectors of indistinguishable particles without classical counterpart.

We note that there is an interesting limit of the above result which corresponds to those configurations where
the overlap integral vanishes (or becomes infinitesimally small):
Z
⇤
hxiab ⌘ dx x a (x) b (x) ! 0. (3.34)

For instance, this would be the case in situations where the one-particle states a (x) and b (x) are only
non-zero in di↵erent regions of x, like when two electrons are really far from each other. This implies that
systems of two identical particles for which the overlap integral Eq. (3.31) can be treated as if they were
distinguishable: it makes full sense to tell two electrons apart if they are separated by kilometers (but not
if the separation is of a few nanometers!).

Page 64 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

The electron spin reloaded. In Sect. 2.5 we discussed how to combine the spin of two particles and to
determine the possible options for the total spin S and its z-component Sz of the system. There we saw
that we can combine the spin of two spin-1/2 fermions (such as electrons) into four states, three of them
characterised by a total spin s = 1,
↵
|1 1i = "" (s = 1, m = 1) ,
1 ↵ ↵
|1 0i = p "# + #" (s = 1, m = 0), (3.35)
2
↵
|1 1i = ## (s = 1, m = 1) .

which are known as the triplet states, and another with vanishing total spin s = 0, the singlet state:

1 ↵ ↵
|0 0i = p "# #" .
2

As we discussed ,the triplet states are symmetric under the exchange of two particles, while the singlet one
is instead antisymmetric: if you exchange the two electrons you get

1 ↵ ↵ 1 ↵ ↵
p #" "# = p "# #" = |0 0i . (3.36)
2 2

What is the connection between the spin combinations of a two-particle system and the general requirement
that the wave function of a system of two fermions must be antisymmetric upon their exchange? Well, for a
two electron system, the total wave function is the product of the spatial and the spin parts of the wave
function. For example, we could have
(r1 , r2 )|s mi , (3.37)

with the spin part |s mi being either one of the three triplet states or the singlet state defined above.13
In such a case, the antisymmetry requirement of the total wave function of the two-electron system then
implies:

• If the two electrons occupy the same position state, one has that (r1 , r2 ) = (r2 , r1 ), so then the
spin state must the antisymmetric singlet state to ensure that the total wave function is antisymmetric
upon the exchange of the two particles.

• If the two electrons have associated a symmetric singlet state, then the spatial part of the wave function
will be non-zero only if it is antisymmetric (r1 , r2 ) = (r2 , r1 ) to ensure that the total wave function
is antisymmetric upon the exchange of the two particles.

The Pauli exclusion principle for electrons

We have thus demonstrated that two electrons can occupy the same position state (say, the
same electronic orbital nlm in the hydrogen atom) only if their spin state is the singlet state,
that is, only if the z-component of their spins points in opposite directions. This is why in a given
non-degenerate quantum state we can only accomodate two fermions: provided their spins are in
opposite direction, this is the only configuration that ensures an antisymmetric total wave function.

13 At least, if we want tout state to have well defined s and m values: as you know, any linear superposition of valid spin

states represents also a bona fide state of the system.

Page 65 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

The generalised symmetrisation principle. In the previous discussion we have restricted ourselves
to systems of two non-interacting particles. However, the generalised symmetrisation principle of
quantum mechanics has associated a much stronger requirement: a system of identical particles must be
either symmetric (for bosons) or antisymmetric (for fermions) upon the exchange of any of the particles
that compose it.
Let us denote the state vector of a general two-particle system as |(1, 2)i. Then we define the exchange
operator P̂ as follows
P̂ |(1, 2)i = |(2, 1)i . (3.38)

The square of this operator is the identity operator 1, given that

⇣ ⌘
P̂ 2 |(1, 2)i = P̂ P̂ |(1, 2)i = P̂ |(2, 1)i = |(1, 2)i , (3.39)

from where it follows that the eigenvalues of P̂ are ±1. If two particles are identical, clearly the Hamiltonian
will be invariant under the exchange r1 ! r2 and therefore P̂ and Ĥ commute and represent compatible
observables: h i
Ĥ, P̂ = 0 . (3.40)

Now, recall the generalised Ehrenfest theorem that determined the time evolution of an observable in
terms of its commutation relations with the Hamiltonian operator:
* +
d i Dh iE
b +
b
@O
hOi = Ĥ, O , (3.41)
dt ~ @t

This relation applied to the case of the exchange operator yields

d
hP i = 0 (3.42)
dt

implying that if a quantum system starts in one of the eigenstates of P̂ (that is, it is either symmetric
hP i = 1 or antisymmetric hP i = 1 upon the exchange of the two particles), it will remain this way forever.
In other words, the symmetry properties of a quantum state upon the exchange of two particles are
time-invariant.
The generalised symmetrisation axiom of quantum mechanics tells us that these properties are the
same for n-particle system composed by indistinguishable fermions or bosons. That is, only states which
satisfy

P̂ij |(1, 2, . . . , i, . . . , j, . . . , n)i = |(1, 2, . . . , j, . . . , i, . . . , n)i = ±|(1, 2, . . . , i, . . . , j, . . . , n)i , (3.43)

are allowed in quantum mechanics, where P̂ij is the exchange operator applied to particles i and j of the
system, and the positive sign holds for bosons and the negative one for fermions.
As mentioned above, actually this generalised symmetrisation requirement is not an axiom but a con-
sequences of the principles of relativistic quantum mechanics (the spin-statistics theorem), though this
di↵erence will not have any practical implication for the discussions that we will have in the rest of the
course.

Page 66 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

3.2 Considerations for multi-electron atoms

Equipped with a better understanding of the quantum mechanical consequences and requirements of systems
composed by identical (indistinguishable) particles, we can extend our discussion of the hydrogen atoms to
atoms composed by more than one electron. The starting point is the Hamiltonian operator Ĥ for a
neutral atom with Z protons, A Z neutrons, and Z electrons, which is given by

XZ  2 ✓ ◆ Z
~ 1 Ze2 1X e2
Ĥ = r2j + , (3.44)
j=1
2m 4⇡✏0 rj 2 |rj rk |
j6=k

which is composed by the kinetic energy term for each electron, the attractive potential between each
electron and the positive charge +Ze of the atomic nucleus, and the electric pair-wise repulsion between
electrons. Note that the factor 1/2 in the repulsive term is introduced to avoid double counting. Except for
the hydrogen case (Z = 1), this Hamiltonian is not solvable unless we introduce some approximations such
as perturbation theory, which we will do later in the course.
The simplest possible multi-electron atom is helium (Z = 2), for which the above Hamiltonian reduces
to the following operator:
X2  2 ✓ ◆
~ 1 Ze2 e2
Ĥ = r2j + , (3.45)
j=1
2m 4⇡✏0 rj |r1 r2 |

The most drastic approximation which allows us to solve the corresponding Schroedinger equation is to
completely neglect the electron repulsion. In this case, the system reduces to two non-interacting
particles (since the electrons now experience only the attractive potential from the atomic nucleus) and as
we know from the previous discussion the total two-particle wave function can now be computed exactly in
terms of the one-particle wave functions.
With this approximation, the Hamiltonian of the helium atom simplifies to

X2  2 ✓ ◆
~ 1 Ze2
Ĥ = r2j , (3.46)
j=1
2m 4⇡✏0 rj

which has the same form of the Hamiltonian for a system of non-interacting particles. Therefore, we know
that in this crude approximation the Helium wave function can be factorised into the product of the
one-particle wavefunctions of a hydrogen-like atom, such that

(r1 , r2 ) = nlm (r1 ) n0 l0 m0 (r2 ) , (3.47)

where note that in general the quantum numbers of the two one-particle wave functions will be di↵erent.

Entangled states

However, this is certainly not the only option for the possible electronic wave functions of the Helium
atom: we can have also linear combinations of the one-body eigenstates of the form

3 4
(r1 , r2 ) = nlm (r1 ) n0 l0 m0 (r2 ) + ñl̃m̃ (r1 ) ñ0 l̃0 m̃0 (r2 ) . (3.48)
5 5
Note that these will be entangled states in that they cannot be written as a combination of specific
one-particle states.

Page 67 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

In these entangled states, the two particles are intrinsically, well, entangled: if I measure the energy (or the
angular momentum) of particle 1, I will know for sure what will be the energy (or the angular momentum)
of particle 2 without measuring it, even if the two particles are separated by a very large spatial distance.
Since the energies of hydrogen-like atoms with Z protons scale as Z 2 EnH in terms of the energy levels of
the hydrogen atom, recall Eq. (2.72), we know that in this approximation the total energy of the Helium
atom system will be specified by the two principal quantum numbers n and n0 and given by

En,n0 = 4 (En + En0 ) , (3.49)

in terms of the energies En of the electronic orbitals of the hydrogen atom. The ground state in this
approximation for the Helium atom will now be given by

8 2(r1 +r2 )/a

0 (r1 , r2 ) = 100 (r1 ) 100 (r2 ) = e . (3.50)
⇡a3
To derive this expression, we have used that the ground state wave function of the hydrogen atom is
r
2 r/a 1 1 r/a
100 (r) = R10 (r)Y00 (✓, ) = e = e , (3.51)
a3/2 4⇡ ⇡a3/2

and then used that from the definition of the Bohr radius, Eq. (2.58), I need to replace a by a/Z if instead
of a single proton I have Z protons in the atomic nucleus, as is the case for hydrogen-like atoms.
Since the spatial part of this Helium ground state wave function, Eq. (3.50), is symmetric, 0 (r1 , r2 ) =
0 (r2 , r1 ), then the fact that electrons are fermions implies that the spin part of the wave function must
be antisymmetric, and thus can only be the singlet configuration, with the z component of their spins
pointing in opposite directions. Hence we conclude that the total wave function of the helium atom will be
given by
1 ↵ ↵
0 (r1 , r2 )|0 0i = p 100 (r1 ) 100 (r2 ) "# #" (3.52)
2
in terms of the tensor product of the spatial and spin states of the wavefunctions, in the approximation
where the neglect the electrostatic repulsion between the two electrons.
Neglecting the repulsive interactions between the two electrons is of course not a good approximation in
any sense (note that the strength of the mutual electron repulsion is only half of the e↵ect of the positive
attraction from the nucleus). Below we provide some strategies in which to improve our predictions for the
energies and the wave functions of the Helium atom.

Shielding and the e↵ective charge. In the very crude approximation where we completely neglect the
repulsive interaction between the two electrons, the ground state of Helium has an energy of

E0 = 4 ⇥ 2 ⇥ ( 13.6 eV) = 109 eV , (3.53)

to be compared with the experimentally determined value of E0 = 78.975 eV, which is o↵ by almost 30%.
Of course, this is not unexpected: the electron repulsive interactions is by no means a subleading e↵ect, and
hence it is expected that our crude approximation is not very close to the actual result.
The basic assumption that underlies the orbital approximation that we have used is that the motion
of the two electrons in the system is uncorrelated between them: it is not a↵ected by its mutual repulsive
interaction. However, we know that this is not the case: the mutual repulsion between the two electrons will
modify their motion around the nucleus to some extent. While we cannot compute fully this e↵ect, we can

Page 68 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

account for it in an approximate manner:

The e↵ective electric charge

In multi-electron atoms, one electron feels the average presence of the other electron. For example,
for some configurations the two electrons will have little overlap, while for other configurations they
will be on average closer. So e↵ectively, the total positive electric charge felt by the electron will be
smaller than the value Z from the protons in the nucleus, since the presence of the other electron will
be reducing it. One then defines the e↵ective charge of the atomic nucleus as the one that one
electron experiences due to the smearing induced by the presence of the other electron.

In the case of the helium atom, we will have that this smearing will lead to an e↵ective charge Ze↵ < Z = 2
smaller than the number of protons in the nucleus. This e↵ect is also known as shielding: the positive
charge of the nucleus felt by a given electron is partially shielded by the other electrons that are also orbiting
around it. In this case, this e↵ective electric charge of the nucleus turns out to be Ze↵ = 1.62. Note that in
principle the value of Ze↵ can be di↵erent depending on the specific orbital, but for the He atom with only
two electrons it is the same for all electronic orbitals.
Hence we find that once we account for the e↵ects of shielding via the e↵ective electric charge, the spatial
wave function of the ground state of the Helium atom will be
3
Ze↵ Zeff (r1 +r2 )/a
0 (r1 , r2 ) = 100 (r1 ) 100 (r2 ) = e , (3.54)
⇡a3
with 1  Ze↵  2 quantifying the strength of the shielding e↵ect. Therefore, we have shown how, within
the orbital approximation, one can construct the wave functions for the electronic orbitals of the helium
atom in terms of the wave functions that we derived in the previous chapter for the hydrogen-like atoms.
Furthermore, we know already that s-type orbitals have a higher likelihood to be found close to the
nucleus at r = 0 and therefore the e↵ective charge that they experience will be larger than that of the p-type
orbitals, which on average are further away from the protons in the nucleus. This e↵ect can be observed from
the table below, which compares the value of the e↵ective electric charge Ze↵ of Helium with that associated
to the di↵erent orbitals of Carbon (with Z = 6):

Element Z Orbital Ze↵

He 2 1s 1.6875
C 6 1s 5.6727
2s 3.2166
2p 3.1358

For similar reasons, the value of the shielding constant for d-type orbitals will also be di↵erent than for the
s- and d-type orbitals: an electron in a d-type orbital is on average more likely be found farther from the
nucleus than that of an p-type orbital, and thus one expects a more intense shielding. We can also see that
for a given type of orbital, say a s-type orbital, the shielding constant will be larger for more excited states
with larger n, which on average are found farther from the atomic nucleus.
Therefore, if we denote by n,l the shielding constant associated to an orbital with quantum numbers
(n, l), we will have the following rules of thumb:

• For orbitals with a fixed principal quantum number n, the larger the value of l the higher the value of

Page 69 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

the shielding constant, for example we have that

3,s  3,p  3,l . (3.55)

• For orbitals with a fixed angular quantum number l, the larger the value of the principal quantum
number n the higher the value of the shielding constant, for example we have that

1,s  2,s  3,s . (3.56)

Note also that this is not a fundamental rule and that there are exceptions to the above general principle.
Therefore, we see that in a multi-electron atom the amount of shielding experienced by the electrons
will depend on the orbital that they occupy. For this reason, in order to evaluate for example the
wavelength of photons involved in the transitions between the orbitals of a multi-electron atom, we need to
use the following expression for the electron energy:
2 0⇣ ⌘ 12 3 ⇣ ⌘2 ⇣ ⌘2
(n,`) (n,`) (n,`)
6 me @ Ze↵ e2 1 Z E1H (13.6 eV) ⇥ Ze↵
A 7
e↵
En,` = 4 2 5 2 = = , (3.57)
2~ 4⇡✏0 n n2 n2

Note that now the energy levels depend both on the quantum numbers n and `, rather than only an n as is
(n,`)
the case of the hydrogen-like atoms. The reason of this di↵erence is that now the electric charge Ze↵ will
vary with the orbital, and di↵erent values of ` will have associated di↵erent e↵ective charges.
To summarise, the e↵ective charge approximation allows us to improve our estimates of the energies
and wave functions for multi-electron atoms, as compared to the scenario where we completely neglect the
electrostatic repulsion between their electrons.

3.3 Implications for solid-state structure

The quantum mechanical theory of identical particles in general, and the associated Pauli exclusion principle
in particular, have huge implications for our understanding of the structure of solid-state matter, such as
crystalline solids. Here we will introduce two quantum-mechanical models for the solid state and highlight
the role of quantum mechanical statistics in determining its properties.

3.3.1 The free-electron gas

We can understand a crystalline solid as a three-dimensional system of atoms (or molecules) which occupy
some geometric, periodic lattice structure. The electrons in these atoms can be classified into core electrons,
that to not participate in chemical bonding, and valence electrons, which corresponding to those electrons
occupying the outermost orbitals and that are responsible to establish chemical bonds. These valence
electrons are known to be delocalized: they are not bound to their original atoms, but can rather move freely
across the whole solid, jumping their way across the di↵erent atoms in the crystalline lattice. Furthermore,
these valence electrons are known to form a electronic band, which is a region of the phase space where
the electrons can occupy any of a continuum of energy values (as opposed to the discrete energy spectrum
that characterise the electronic orbitals of individual atoms).
Here, first of all we will study the consequences of the fact that the electrons are delocalised in the solid by
means of the free electron gas model, and afterwards we will quantify the implications of the periodicity

Page 70 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

of the atoms in a crystal to explain the appearance of the electronic energy bands using a periodic potential
composed by Dirac delta functions.
To very first approximation, we can therefore consider a solid as an infinite square potential well in
three dimensions: an electron is free to move everywhere within the solid but cannot jump outside it. In
other words, the potential that the electron will experience in this system is given by

V (x, y, z) = 0 0 < x < l x , 0 < y < ly , 0 < z < l z (3.58)

V (x, y, z) = 1 otherwise

where lx , ly , lz are the dimensions of the solid (the potential well) across the x, y, and z directions. For
this potential, the Schroedinger equation is nothing but the free-particle Schroedinger equation in three
dimensions. Its solutions will be subject to the usual boundary condition that the wave function must
vanish at the surface of the solid, in the same way as in the infinite quantum well in one dimension:

(x = 0, y, z) = (x = lx , y, z) = (x, y = 0, z) = (x, y = ly , z) = (x, y, z = 0) = (x, y, z = lz ) = 0 .

Since we have a free-particle time of equation, It is then to convince yourselves that its wave function
factorises into a product of the one-particle solutions of the infinite quantum well,

(x, y, z) = X(x)Y (y)Z(z) , (3.59)

where we have defined the individual components of the wave function as:
r
2mEx
X(x) = Ax cos(kx x) + Bx sin(kx x) , , kx =
~2
r
2mEy
Y (y) = Ay cos(ky y) + By sin(kx y) , ky = ,
~2
r
2mEz
Z(z) = Az cos(kz z) + Bz sin(kx z) , kz = ,
~2
which satisfy the boundary conditions if Ax = Ay = Az = 0 and kx lx = nx ⇡, ky ly = ny ⇡ and kz lz = nz ⇡
where as usual ni are positive integers and Bi are integration constants fixed by the normalisation of the
wave function.
Therefore we have demonstrated that the wave function for an electron confined into a three-dimensional
box (our model for a solid) is given by
s ✓ ◆ ✓ ◆ ✓ ◆
8 nx ⇡ ny ⇡ nz ⇡
nx ny nz (x, y, z) = sin x sin y sin z , (3.60)
lx ly lz lx ly lz

while the corresponding electron energies will be

!
~2 ⇡ 2 n2x n2y n2z ~2 k 2
E n1 n 2 n 3 = + + = , (3.61)
2m lx2 ly2 lz2 2m

in terms of the magnitude of the wave vector k = (kx , ky , kz ). Every possible value of the wave vector

Page 71 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

corresponds to a di↵erent choice of quantum numbers:

✓ ◆
⇡nx ⇡ny ⇡nz
(kx , ky , kz ) = , , , nx , ny , nz = 1, 2, 3, . . . , (3.62)
lx ly lz

and therefore to a di↵erent electronic wavefunction. Note that the wave vector has units of 1/length, and it
is often referred as to spanning the conjugate space of position space.
In order to understand the properties of this free-electron gas system, it is useful to think in terms of
the three-dimensional vector space spanned by the wave vector k = (kx , ky , kz ). If you imagine a grid at
integer values of ki , then every node in the grid represents a possible one-particle state of the system.
Since each grid point is separated by ⇡/lx , ⇡/ly , and ⇡/lz in the x, y, and z directions respectively, the
volume of each block in the grid is
⇡3 ⇡3
Vk block = = . (3.63)
lx l y lz V
In these discussions one must be careful, since Vk block refers to a volume in the k-space (the conjugate
space), rather than V which is volume in position space.
If we had a single electron in this system, we would be done and we could end the discussion here. But
of course, a solid is not composed by a single electron but by a very large number of them. For example,
for a solid composed by N atoms, each atom will contribute with handful d of valence electrons, and hence
the total number of electrons that will compose our free electron gas (free, except for not being able to
jump outside the boundaries of the material) will be

Nelectrons = N d . (3.64)

Since electrons are fermions, they must obey the Pauli exclusion principle, and hence for a given value of wave
vector k = (kx , ky , kz ) (hence, for a given volume Vk block = ⇡ 3 /V ) we can have at most two electrons.
These electrons must form a singlet state with spins pointing in opposite directions.
Therefore, the total volume in k-space occupied by the electrons in this system will be

1 1 ⇡3
⇥ Vk block ⇥ Nelectrons = ⇥ ⇥ Nd , (3.65)
2 2 V
where the 1/2 factor comes since two electrons occupy one Vk block . Since (kx , ky , kz ) is an array of positive
integer numbers, electrons will occupy first the states with lowest ki and once these are occupied they will
occupy higher wave vectors. The volume of the resulting octant of a sphere in k-space, once all electrons
have filled the available states, will be given by
✓ ◆ ✓ ◆
1 4 3 Nd ⇡3
⇡k = , (3.66)
8 3 F 2 V

where the radius of this k-space sphere is known as the Fermi momentum:
✓ ◆1/3
3N d⇡ 2
kF = = (3⇢⇡ 2 )1/3 (3.67)
V

where we have defined the free electron density as ⇢ ⌘ N d/V , namely the number of free (valence)
electrons in the free electron gas system per (physical) unit volume. The surface of this sphere, populated
by the electrons with the highest values of the wave vector k, is known as the Fermi surface.
The electrons with wave vector k spanning the surface of the Fermi sphere will have the highest energies

Page 72 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

in the whole material. By recalling the dispersion relation for free particles, E = ~2 k 2 /2m, we can determine
the energy of those electrons, which is denoted as the Fermi energy:

~2 kF2 ~2
EF = = (3⇢⇡ 2 )2/3 (3.68)
2m 2m
The position of the Fermi energy has a crucial impact of the properties of a solid, for example in determining
its electrical conductivity properties. From this result we observe how the denser the material, the higher
the value of its Fermi energy.

Position versus moment (wave number) space

You should be careful with the distinction between position and moment space, also denoted as x-
space and k-space. The electrons occupy only a sphere (octant) in moment (that is, in energy space),
but they remain delocalised in position space. Each electron, irrespectively of its wave vector k, has
some finite probability of being found everywhere within the solid, as indicated by the wave function
Eq. (3.60). Hence the Fermi sphere is a geometrical construct in moment space, which does not have
a counterpart in position space.

We can now evaluate the total energy of my system of N d electrons in this free-electron gas model. Within
the Fermi sphere in k-space, a shell with (infinitesimal) thickness dk contains a volume

1
dV = 4⇡k 2 dk . (3.69)
8

Now, the density of electrons in k-space is 2/Vk block = 2V /⇡ 3 , since each k-block within the Fermi
sphere contains two electrons. Hence, we have that the number of electrons in this infinitesimal shell is

1 2V V
dNel = 4⇡k 2 dk ⇥ 3 = 2 k 2 dk . (3.70)
8 ⇡ ⇡

Furthermore, a state represented by the wave vector k has associated an energy of E = ~2 k 2 /2m, and
therefore the energy of the shell will be
✓ ◆ ✓ 2 2◆
~2 k 2 V 2 ~ k V ~2 k 4
dE = dNel ⇥ = k dk ⇥ = dk . (3.71)
2m ⇡2 2m 2⇡ 2 m

At this point, in order to evaluate the total energy of the free-electron gas system, I only need to integrate
over all available wave vectors, from k ' 0 for the ground state up to the Fermi energy kF . We obtain the
following result:
Z kF
V ~2 k 4 ~2 kF5 V
Etot = 2
dk = . (3.72)
0 2⇡ m 10⇡ 2 m
It is convenient to simplify this important result by exploiting the expression of the Fermi energy Eq. (3.68),
which remember is the energy of the electrons with the highest wave numbers k of the system. By doing
this we find that the total energy of the free electron gas is:

~2 (3⇡ 2 N d)5/3 2/3

Etot = V . (3.73)
10⇡ 2 m
5/3
The total energy grows with the number of free electrons as Nel , and decreases with the volume of the solid.
The latter result can be understood by the fact that in the V ! 1 limit and fixed number of electrons, the

Page 73 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

volume of the k-block vanishes and electrons occupy states with very low energies.
While we are referring of this system as a free-electron gas, this gas is very di↵erent from the usual gases
in that the electrons themselves are non-interacting, and thus their energy Eq. (3.73) is purely quantum
mechanical in origin (as opposed to a regular gas, whose energy is ultimately thermal in origin). Despite
this di↵erent origin, this quantum mechanical energy exerts a very measurable pressure on the walls of a
solid, in the same way as the pressure of a gas arises from the collisions of its constituent molecules with the
walls of the contained.
We can determine the quantum mechanical pressure exerted by our free-electron gas as follows. If the
solid (container of the electron gas) expands by an amount dV , you can see that its total energy decreases:

2 ~2 (3⇡ 2 N d)5/3 5/3 2 dV

dEtot = V dV = Etot . (3.74)
3 10⇡ 2 m 3 V
If you recall from your thermal physics course the relation between work, pressure and volume,

dW = P dV , (3.75)

we find that the quantum mechanical pressure exerted by the electrons on the walls of the solid is given by:

2 Etot (3⇡ 2 )2/3 ~2 5/3

P = = ⇢ , (3.76)
3 V 5m
which is stronger the higher the electron density of the material.
The existence of such quantum mechanical pressure, sometimes referred to as degeneracy pressure,
provides some input to the question of why a cold solid does not collapse, demonstrating that what
(partially) ensures its “solidity” is not thermal e↵ects (which have been ignored) nor the interactions of the
electrons among them or with the atomic nuclei (which we have neglected). A cold solid does not collapse
because it is composed of identical fermions, and fermions do not like to occupy the same quantum state!
Note that it is quite remarkable how we have managed to produce a reasonably good prediction for the
properties of a solid using exclusively the fact that electrons are fermions, without saying a single word
about their interactions. This demonstrates that many of the fundamental properties of solids are dictated
by basic quantum mechanical considerations, rather than from the details of the specific interactions present
within in the system.

3.3.2 Band structure in solids

The main limitation of the free electron gas model is that it considers a solid as a collection of non-interacting
electrons restricted to the material volume, but ignores the fact that these electrons arise from the valence
orbitals of atoms that are distributed in a more or less regular manner within the solid. Thus this model
neglects the interactions of these valence electrons with the atoms that compose the solid. In order to
improve the free electron model, we need to account for, at least in an approximate way, the interactions
between the atoms in the solid and the electrons.
As in the case of the free-electron gas, it can be shown that the specific details of the atom-electron
interactions are not relevant, and that many important properties of solids can be recovered just by taking
into account the fact that the atomic nuclei are stationary and distributed regularly within the solid.
Hence, the defining feature of the model we are going to present now is that the regular arrangement of the
atoms within a crystalline solid gives rise to periodic potential, which repeats itself in the three spatial
directions.

Page 74 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

The discussion of quantum systems characterised by periodic potentials is greatly simplified by the
powerful Bloch’s Theorem. Consider a one-dimensional quantum system characterised by a periodic
potential, namely a potential that repeats itself after a distance a:

V (x + a) = V (x) , (3.77)

which of course implies that V (x + 2a) = V (x + a) = V (x) or in general that

V (x + na) = V (x) for n = 1, 2, 3, . . . . (3.78)

The wave function for a particle propagating through this potential energy will satisfy the one-dimensional
Schroedinger equation,
~2 d 2
(x) + V (x) (x) = E (x) . (3.79)
2m dx2
Bloch’s Theorem tells us that, given the periodic nature of the potential defined by Eq. (3.78), the solution
of the one-dimensional Schroedinger equation, must satisfy:

(x + a) = eiqa (x) , (3.80)

with q some constant number independent of the position x. To demonstrate this property, note that if (x)
satisfies Eq. (3.79) then it will also satisfy

~2 d 2
(x + a) + V (x + a) (x + a) = E (x + a) , (3.81)
2m dx2
where I have rescaled x ! x + a. But we are dealing with a periodic potential as indicated by the
periodicity condition Eq. (3.78), and hence:

~2 d 2
(x + a) + V (x) (x + a) = E (x + a) , (3.82)
2m dx2
and if now we replace
(x + a) = eiqa (x) (3.83)

we reproduce the original equation Eq. (3.79), demonstrating Bloch’s Theorem. The overall prefactor eiqa
does not depend on x and thus cancels out in the Schroedinger equation.

Periodicity of the electron probability distribution

For the time being q is a general complex number, and thus it could change the normalisation of the
wave function. In short we will demonstrate that q 2 R and thus the eiqa prefactor is a complex
phase. This is a reassuring result, given the periodicity of the potential heavily suggests that the
probability distribution associated to the electron should also be periodic, and hence that

| (x + a)|2 = | (x)|2 , (3.84)

which is indeed the case thanks to Bloch’s Theorem, Eq. (3.80), given that eiqa is a complex phase.
Therefore in a crystalline solid with a periodic potential the probability distribution associated to its
valence electrons will be periodic as well.

Page 75 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

While any real solid is finite, finite-size e↵ects associated to its boundaries can be neglected for macroscopic
solids as compared to the dominant bulk (volume) e↵ects. We can ignore them formally by imposing the
periodic boundary condition
(x + N a) = (x) , (3.85)

where N = O 1023 is a macroscopic number of periods, which corresponds to a macroscopic number of

atoms. Taking into account the periodicity properties of the wave function, Eq. (3.80), we find that imposing
this periodic boundary conditions leads to

(x + N a) = eiqN a (x) = (x) ! qN a = 2⇡n , (3.86)

and thus the constant q is real and quantised:

2⇡n
qn = , n = 0, ±1, ±2, ±3, . . . (3.87)
Na

Note that (1/N a) ' 10 13 for a typical interatomic separation of a = O 10 10 , and hence the values
of qn will form a quasi-continuum. Note that if we had not imposed the periodic boundary conditions
then q would indeed take continuous values, but the di↵erence between quasi-continuum actual continuum
is irrelevant in practice.

The usefulness of Bloch’s Theorem

The crucial relevance of Bloch’s theorem is that we have reduced the problem of solving the
Schroedinger equation on the whole solid to that of solving it in a unit cell, for example the
cell defined by 0  x  a. Once the wave function for this unit cell is known, then Eq. (3.80)
guarantees that we can evaluate it for any other cell within the solid.

Armed with the powerful Bloch Theorem, we can now present our model for the band structure of solids.
Still restricting ourselves to the one-dimensional case, let us assume that our periodic potential is a linear
combination of Dirac delta functions:
N
X1
V (x) = ↵ (x ja) . (3.88)
j=0

That is, we have infinite spikes localised at the nodes x = ja of the lattice with j being an integer number.
The specific details of the potential are not crucial (one could have used finite square barrier potentials,
for example), the really important property is their periodicity. Indeed, while this potential is far from
the real physical situation (we should have Coulomb-type attractive interactions between the electrons and
the positive atomic cores), it allows us to evaluate the physical consequences of a periodic potential in the
resulting properties of solids.
The potential energy defined by Eq. (3.88) is clearly a periodic potential, since V (x + a) = V (x), as you
can check as follows
N
X1 NX1
V (x + a) = ↵ (x ja + a) = ↵ (x (j 1)a) , (3.89)
j=0 j=0

Page 76 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

but now you can redefine j 0 = j 1 to have

N
X1 N
X2
V (x + a) = ↵ (x (j 1)a) = ↵ (x j 0 a) = V (x) (3.90)
j=0 j0= 1

since we are imposing periodic boundary conditions for the potential, Eq. (3.85) such that the lattice
node at j 0 = 1 in the sum is identified with the node with j 0 = N 1. Indeed, is we plug in x = a in
Eq. (3.85) we get

( a + N a) = ( a) ! j0 = N 1 identified with j 0 = 1, (3.91)

and hence the potential is indeed periodic as required.

By virtue of Bloch’s Theorem, we only need to solve the Schroedinger equation for this potential in a
single “crystal” cell, which we take to be 0  x  a. The delta function potential Eq. (3.88) is clearly zero
for 0 < x < a, and it is only non-zero for x = 0 and x = a. Therefore, for 0  x  a we recover the usual
free-particle wave functions
p
(x) = A sin(kx) + B cos(kx) , k= 2mE/~ , 0 < x < a. (3.92)

We can now use Bloch’s Theorem,

(x + a) = eiqa (x) ! (x) = e iqa

(x + a) , (3.93)

to compute the wave function in the cell to the left of this one, a < x < 0, and we find

iqa
p
(x) = e (A sin(k(x + a)) + B cos(k(x + a)) , k= 2mE/~ , a < x < 0. (3.94)

In order to determine the integration constants A and B, we need to impose the relevant local boundary
conditions on the wave function of this system. We will this way find the conditions to be satisfied by the
physical wave functions of this periodic system.

• The wave function must be continuous everywhere, and hence for x = 0 we must impose:

iqa
B=e (A sin(ka)) + B cos(ka)) (3.95)

which is a transcendental equation similar to the one that arises in the finite potential barrier and
related problems (which can only be solved numerically).

• The derivative of the wave function is discontinuous across infinite potentials, such as the Dirac
delta function of this system, with the value of the discontinuity given by
Z ✏
d (x) d (x) 2m
= dx V (x) (x) . (3.96)
dx dx ~2 ✏
+✏ x= ✏

where I am assuming that the potential V (x) is infinite at x0 = 0. Our potential at x = 0 is a delta
function and hence the RHS reads
Z Z
2m ✏ 2m↵ ✏ 2m↵ 2m↵
dx V (x) (x) = 2 dx (x) (x) = 2 (0) = 2 B . (3.97)
~2 ✏ ~ ✏ ~ ~

Page 77 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

Now in the original cell, x = +✏, the derivative is given by

d (x) d (x)
= Ak cos(kx) Bk sin(kx) , = kA (3.98)
dx dx
+✏

while instead for x = ✏ the corresponding derivative is evaluated to be:

d (x) iqa
= e (Ak cos(k(x + a)) Bk sin(k(x + a)) ,
dx
d (x) iqa
= e (Ak cos(ka)) Bk sin(ka)) (3.99)
dx
✏

and thus the boundary condition for the derivate of the wave function reads:

⇥ iqa
⇤ 2m↵
kA e k (A cos(ka)) B sin(ka)) = 2 B . (3.100)
~

Eqns. (3.95) and (3.100) provide two relations that constrain the integration constants A and B and the
allowed values of the energy k of the particles propagating in this periodic potential.
To solve this system of equations, we can first use Eq. (3.95) to determine A in terms of B

eiqa cos(ka)
A=B (3.101)
sin(ka)

and then insert this relation into Eq. (3.100), where the dependence on B cancels out. This indicates that B
is a normalisation constant fixed by the wave function normalisation requirement. Doing some algebra, we
end up with the following relation between the quasi-continuous parameter q and the particle wave vector k
as a function of the strength ↵ of the potential:
m↵
cos(qa) = cos(ka) + sin(ka) . (3.102)
~2 k
This relation illustrates how for given values of the potential strengths ↵ and of the lattice parameter a
the energies of the electrons E = ~2 k 2 /2m in the system are related to the values of q = 2⇡n/N a, and hence
they will be quantised. Indeed, for given values of q, a. and ↵, only electron energies such that

~2 k 2 m↵
E= satisfying cos(qa) = cos(ka) + sin(ka) , (3.103)
2m ~2 k
are allowed solutions of the Schroedinger equation of this periodic potential. However, there is now a marked
di↵erence with previous systems that we have considered: the fact that q is essentially a continuous variable
(since N is macroscopic) implies that instead of discrete energy levels, we will have continuous energy
bands separated by gaps without any allowed states.
To determine which are the allowed values of the energy E of the electrons in this system, we can write
z = ka and = ma↵/~2 such that Eq. (3.102) becomes
✓ ◆
2⇡n
cos = cos(z) + sin(z) = f (z) . (3.104)
N z

For a given value of , which is proportional to the strength of the interaction between the electrons and the
atomic nuclei, for each value of n = 1, 2, 3, 4, . . . there will be a solution for z.

Page 78 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

However, N is a huge number, so for any value of cos(qa) between -1 and +1 it is essentially guaranteed
that we will find a value of n that fulfills this condition. Hence, any value of k = z/a such that |f (z)|  1
corresponds to an allowed energy state for my system: we therefore find that the valence electrons
moving in the periodic potential of a solid cluster into bands of allowed energy separated by energy gaps
(for |f (z)| > 1). This is a rather significant di↵erence as compared to e.g. the hydrogen atom, or to any other
quantum systems: in periodic potentials, the allowed levels of the electron energies cluster into continuum
bands separated by finite gaps where no solution is possible.
Fig. 3.1 displays the function f (z), Eq. (3.104) as a function of z for = 30 and = 4. We indicate
with dashed horizontal lines the region |f (z)|  1 which defines the allowed electronic bands. In the bottom
panels, we indicate with a solid grey rectangle the location of these allowed electronic bands in the two
scenarios. Note how the width of the band increases as z is increased, and eventually becomes a continuum
without any gaps breaking the energy band. Note also that smaller the value of , the earlier the energy
continuum kicks in (since in ! 0 limit the strength of the periodic potential vanishes).
Note also that in the limit ! 0 we can always find a solution to Eq. (3.104): this means that in this
limit the energies are not quantised and can take any possible value. This is not unexpected, since for ! 0
we recover the free particle system (subject to periodic boundary condition) and there we know that any
positive value of the energy is physically allowed.

Summary
We can now recapitulate what have we learned in this chapter concerning the quantum mechanics for systems
composed by identical particles:

I/ Quantum mechanics deals with truly indistinguishable particles: an electron is identical in all respects
to every other electron in the Universe.

II/ Systems composed by identical particles in quantum mechanics behave in a very di↵erent way as
compared to systems composed by distinguishable particles.

III/ Depending on their spin, particles can be classified into fermions and bosons. Bosons behave as if
they experienced some attractive interaction and tend to cluster together in the same quantum state.
Fermions, on the contrary, behave as if they experienced some repulsive interaction and tend to get far
for each other.

IV/ The Pauli exclusion principle for fermions tells us that a given quantum state can be occupied at most
by one fermion at the same time.

V/ The free electron gas model allows us to explain some important properties of solids, and shows how
there exists a quantum mechanical pressure from the fermion degeneracy that partially explains the
stability of solids.

VI/ The existence of continuous energy bands for the valence electrons in solids can be explained by the
presence of a periodic potential, where the electrons move, generated by the atomic cores sitting at
regularly-spaced locations within the crystalline lattice of a solid.

Page 79 of 92
Dr Juan Rojo Quantum Mechanics 2: Lecture Notes February 23, 2021

Figure 3.1: The function f (z), Eq. (3.104) as a function of z for = 30 and = 4. We indicate with dashed
horizontal lines the region |f (z)|  1 which defines the allowed electronic bands. In the bottom panels, we indicate
with a solid grey rectangle the location of these allowed electronic bands in the two scenarios. Note how the width
of the band increases as z is increased, and eventually becomes a continuum without any gaps breaking the energy
band. The smaller the value of , the earlier the energy continuum kicks in.

Page 80 of 92